Skip to main content

基于Raspberry Pi 5的Frigate NVR

介绍​

Frigate NVR 是一个开源网络视频录像机，专为使用AI模型进行实时目标检测而设计。与 Raspberry Pi 5 配合使用，可在边缘实现高效的视频监控。本指南将引导您完成安装和配置过程，以获得最佳设置。

官方 Frigate​

准备硬件​

推荐组件​

reComputer AI R2130
立即获取 🖱️

另外，您还需要至少一个摄像头用于视频流传输。您可以参考推荐硬件查看推荐的摄像头。

安装Hailo PCIe驱动​

步骤1：启用PCIe Gen 3​

打开终端并运行以下命令。

sudo apt update
sudo raspi-config

在对话框中，选择 6 Advanced Options，然后选择 A8 PCIe Speed。

6 Advanced Options A8 PCIe Speed

选择"Yes"以启用 PCIe Gen 3 模式。

Choose Yes

之后，点击"Finish"退出。

编辑 /boot/firmware/config.txt，在文件末尾添加以下行。

dtoverlay=pciex1-compat-pi5,no-mip

步骤 2：安装 Hailo PCIe 驱动程序​

安装 dkms 包。安装 Hailo PCIe 驱动程序需要此包。

sudo apt update
sudo apt install dkms

Get Hailo PCIe Driver from GitHub.

git clone https://github.com/hailo-ai/hailort-drivers
cd hailort-drivers/linux/pcie
git checkout 24e7ff2fb58fab7029024c1a1d3f2d1914f56d7b

Then, install the Hailo PCIe driver.

sudo make install_dkms

After installation, download firmware for Hailo and copy it to the /lib/firmware/hailo directory.

cd ../..
./download_firmware.sh
sudo mkdir -p /lib/firmware/hailo
sudo cp hailo8_fw*.bin /lib/firmware/hailo/hailo8_fw.bin

为了避免 PCIe max_desc_page_size 问题，我们还需要在 /etc/modprobe.d/hailo_pci.conf 中创建一个规则，内容如下。

options hailo_pci force_desc_page_size=4096

Restart the system to take effect.

sudo reboot

系统重启后，Hailo PCIe 驱动程序安装成功。将创建 /dev/hailo0 设备。通过运行以下命令检查设备。

$ ls /dev/hailo*
/dev/hailo0

安装 Frigate NVR​

在这一部分，我们假设您已经设置好摄像头并准备好使用 RTSP 协议以 1920x1080 分辨率进行流媒体传输。

  • 示例 RTSP URL：rtsp://admin:[email protected]:554/cam/realmonitor?channel=1&subtype=0

步骤 1：准备环境​

  1. 更新您的系统：

    sudo apt update

  2. 安装 Docker：

    curl -fsSL get.docker.com | bash
    sudo usermod -aG docker $USER

  3. Reboot the system:

    sudo reboot

步骤 2：部署 Frigate​

  1. 拉取 Frigate 镜像：

    前往 Package frigate，选择一个带有 -h8l 后缀的镜像。在这个例子中，我们选择 ghcr.io/blakeblackshear/frigate:0.15.0-rc2-h8l。

    docker pull ghcr.io/blakeblackshear/frigate:0.15.0-rc2-h8l

  2. 创建 Docker Compose 文件：

    这是 frigate.yml 文件的示例，hailo0 设备是您在上一步中创建的设备，配置文件位于 ./config 目录中，数据文件位于 ./data 目录中：

    services:
        frigate-hailo:
            container_name: frigate-hailo
            privileged: true
            restart: unless-stopped
            image: ghcr.io/blakeblackshear/frigate:0.15.0-rc2-h8l
            shm_size: 1024mb
            devices:
                - /dev/hailo0:/dev/hailo0
            volumes:
                - /etc/localtime:/etc/localtime:ro
                - ./config/:/config
                - ./data/db/:/data/db
                - ./data/storage:/media/frigate
                - type: tmpfs
                  target: /tmp/cache
                  tmpfs:
                    size: 1g
            ports:
                - 5000:5000

  3. 编辑 Frigate 配置：

    这是 config/config.yml 文件的示例，用于 Frigate 应用程序：

    database:
        path: /data/db/frigate.db

    go2rtc:
        streams:
            home:
                - rtsp://admin:[email protected]:554/cam/realmonitor?channel=1&subtype=0

    cameras:
        home:
            ffmpeg:
              inputs:
                - path: rtsp://admin:[email protected]:554/cam/realmonitor?channel=1&subtype=0
                  roles:
                    - record
                    - detect

    mqtt:
        enabled: False

    objects:
        track:
            - person
            - cat

    detectors:
        hailo8l:
            type: hailo8l
            device: PCIe

    model:
        width: 300
        height: 300
        model_type: ssd
        path: /config/model_cache/h8l_cache/ssd_mobilenet_v1.hef

    version: 0.15-1

  4. 启动 Docker 实例：

    docker compose -f frigate.yml up -d

    Frigate 启动并运行后，您可以通过 http://<your-raspberry-pi-ip>:5000 访问 Frigate Web UI 来检查摄像头流。

    frigate-web

Seeed Frigate​

修改 PCIe 设置​

打开 config.txt

sudo nano /boot/firmware/config.txt

Add the following text to config.txt

dtparam=pciex1_gen=3
dtoverlay=pciex1-compat-pi5,no-mip

Then use Ctrl+x to save the file and reboot the AI box

Install docker and hailo-all​

sudo apt update
sudo apt install hailo-all
curl -fsSL https://get.docker.com -o get-docker.sh
sudo sh get-docker.sh
sudo usermod -aG docker $USER
sudo apt install docker-compose-plugin

Create yml file​

Create frigate.yml for docker compose to run frigate

cd ~
sudo nano frigate.yml

Here is an example of frigate.yml

version: "3.9"
services:
    frigate-hailo:
        container_name: frigate-hailo
        privileged: true
        restart: unless-stopped
        image: mjqx2023/frigate_seeed:latest
        shm_size: 1024mb
        environment:
            - HAILO_MONITOR=1
        devices:
            - /dev/hailo0:/dev/hailo0
        volumes:
            - /etc/localtime:/etc/localtime:ro
            - ./config/:/config
            - ./data/db/:/data/db
            - ./data/storage:/media/frigate
            - type: tmpfs
              target: /tmp/cache
              tmpfs:
                  size: 1000000000
        ports:
            - 5000:5000
            - 1984:1984
            - 5002:5002
            - 5003:5003

Download yolo model and create config.yml.

mkdir config && cd config && mkdir model_cache
cd model_cache && wget https://hailo-model-zoo.s3.eu-west-2.amazonaws.com/ModelZoo/Compiled/v2.14.0/hailo8/yolov8n.hef

cd .. && nano config.yml

Here is an example of config.yml

database:
  path: /data/db/frigate.db

go2rtc:
  streams:
    home:
      - rtsp://admin:[email protected]:554/cam/realmonitor?channel=1&subtype=1
    home1:
      - rtsp://admin:[email protected]:554/cam/realmonitor?channel=1&subtype=1
    yard:
      - rtsp://admin:[email protected]:554/cam/realmonitor?channel=1&subtype=1
    yard1:
      - rtsp://admin:[email protected]:554/cam/realmonitor?channel=1&subtype=1

cameras:
  home:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/home
          roles:
            - detect

    motion:
      threshold: 45
      contour_area: 60
      improve_contrast: true

  yard:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/yard
          roles:
            - detect

    motion:
      threshold: 45
      contour_area: 60
      improve_contrast: true
      
  home1:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/home1
          roles:
            - detect


    motion:
      threshold: 45
      contour_area: 60
      improve_contrast: true
      
  yard1:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/yard1
          roles:
            - detect

    motion:
      threshold: 45
      contour_area: 60
      improve_contrast: true
mqtt:
  enabled: false

objects:
  track:
    - person
    - cat

detectors:
  hailo8l:
    type: hailo8l
    device: PCIe

model:
  width: 640
  height: 640
  path: /config/model_cache/yolov8n.hef

version: 0.16-0

camera_groups:
  test:
    order: 1
    icon: LuAlignEndHorizontal
    cameras:
      - home
      - home1
      - yard
      - yard1

Pull docker image and run frigate​

Pull frigate_seeed image

cd ~
docker pull mjqx2023/frigate_seeed
docker compose -f frigate.yml start

然后在您的 AI 盒子网页浏览器中打开 localhost:5000： frigate-web

技术支持与产品讨论​

感谢您选择我们的产品！我们在这里为您提供不同的支持，以确保您使用我们产品的体验尽可能顺畅。我们提供多种沟通渠道，以满足不同的偏好和需求。

Loading Comments...